In C language:
#define GROUP_SIZE 8
More seriously:
Until now it was the size of a group which is the list of nodes that are the closest to an XOR address. These are the nodes that manage the data having this address, so GROUP_SIZE is also the number of times the data is stored in the network (= duplication factor).
But with new routing, GROUP_SIZE is also the number of elders in a section.
I am not familiar enough with the routing inner workings to see why these 2 numbers are the same. Besides, I don’t think that the elders belong necessarily to a group because they can be far apart in the section, so calling their number GROUP_SIZE is odd. Maybe another name should be chosen to convey what these 2 numbers have in common.